NYC Rule Would Require Chatbots to Offer a Way to Reach a Human

Businesses that serve New Yorkers through chatbots or automated phone menus would have to tell customers they are talking to a machine and offer a route to a human, under rules the Department of Consumer and Worker Protection proposed in a city notice published Oct. 20. A public hearing is set for Nov. 19.

What the rules would require

• Disclosure. At the start of an automated interaction, in the language being used, a business must clearly say the customer is dealing with an automated system and not a person.
• No dead-end loops. It would be deceptive to promise customer service and then repeatedly send people to menus that do not address their question, or answer with “unresponsive, irrelevant, or erroneous information.”
• A path to a person. Bot-only service is presumed deceptive unless the business offers a one-step transfer such as “Press 0” or “Say agent,” with an estimated wait. If no human is available, it must point to an alternative with hours or a response time, such as a phone line, live chat, callback, contact form, text line or email.
• Honest contact pages. Labeling a phone number or email “customer service” or “contact us” when no service is offered there would be deceptive.
• Records. Businesses using these systems must keep compliance records for at least three years.

One new section applies to every rule the agency enforces, not just customer service: it is no defense that “a technology” made the claim, left out the information or committed the violation. The agency says that has always been the law and the rule “makes this principle explicit.”

Penalties: $525 to $3,500

A new penalty schedule sets fines of $525 for a first violation, $1,050 for a second and $3,500 for a third or later violation within two years. The same amounts apply to failing to disclose a bot, failing to provide promised service, offering no service on a listed channel and failing to keep records.

Each provision cited in a violation counts as a separate violation.

The agency’s case

The notice cites a Gartner finding that chatbot resolution rates range from 17% for billing disputes to 58% for returns or cancellations. It also cites a Twilio survey in which 40% of consumers said AI repeats itself or gets stuck in loops.

The agency says the rule was not on its regulatory agenda this year because it “was not contemplated” when the agenda was published.

Also from DCWP: debt collection rule pushed to 2027

In a separate notice published Oct. 21, the agency formally adopted a delay of its SHIELD debt collection rule from Sept. 1, 2026, to Jan. 1, 2027. The rule covers validation notices, disputes and debt verification.

The agency says the delay gives it time to answer questions and gives the industry time to make operational changes. It received one comment, from an industry representative, who supported the delay.

What the notices do not say

The chatbot proposal does not define how long a queue must be to count as “unreasonably long.”

Related coverage

• NYC Wants to Make the Price You See the Price You Pay

Details as filed. Verify times and locations before attending.